Pitchalytics

Pitchalytics leverages dynamic algorithms to allow baseball teams to take in a baseball situation and see which pitch will be the most effective as well as generating various analytical reports.

Comment

Inspiration

Pitching in baseball is always a mind game. Pitchers and catchers work together to build game plans and call the pitches they think will fool the hitter in any given situation. Watching baseball growing up, this was my favorite part of the game: pitch selection. So now we wonder, how much more effective can pitch calling be if that pitch selection was backed by data?

What it does

The app has an input for pitch type and result of pitch and automatically updates the game situation. It then uses an AI algorithm analyzing all past data to suggest the next pitch type to throw based on the stored previous pitch(s) and the newly generated game situation. It also generates reports for what two pitch and three pitch sequences work best.

How we built it

We used advanced data analytics to weight scores of past pitches based on the similarity of those pitches to the current situation, specifically looking at sequences of pitches. The algorithm combs through all the past data from a pitcher in a dataset and using pandas and Machine Learning techniques to optimize pitch selection. We utilized React in order to develop a frontend that keeps track of the game or practice while loading recommendations along the way. To deploy this application, we took advantage of the services provided by AWS. We used AWS CloudFormation to make deployments easy, along with AWS API Gateway and AWS Lambda to deploy the backend algorithms. We also used AWS S3 and AWS CloudFront to host the frontend of the website.

Challenges we ran into

Most of the challenges were normal hackathon problems: debugging code, making the the idea applicable, and finding useable data

Accomplishments that I'm proud of

We think studying pitch type sequencing through an analytical lens is a groundbreaking idea. The algorithm is strong and the front end of the app is very attractive and it is very easy to use.

What we learned

We learned some powerful techniques in app building, and from talking with pitching coach Danny Borrell, we were able to learn about some important tools and available data that would be useful in expanding the strength of our product

What's next for Pitchalytics

We'd love to apply Georgia Tech baseball data to the algorithm and build out the app to help them win games. This app is useful at all levels of baseball and can be a powerful tool for any interested team.

Built With

Share this project:

Updates